Recording medium, information processing method, and information processing apparatus

Abstract

A non-transitory computer-readable recording medium storing an information processing program that causes at least one computer to execute a process, the process includes, acquiring a matrix forming a certain matrix equation; dividing the matrix into a plurality of blocks each including an element having a certain attribute based on an attribute of each element of a plurality of elements included in the matrix; changing a scale of each element of the plurality of elements included in each block of the plurality of blocks; and setting the matrix after the scale is changed to a target to be inputted to a machine learning model that performs a matrix operation.

What is claimed is: 
     
         1 . A non-transitory computer-readable recording medium storing an information processing program that causes at least one computer to execute a process, the process comprising:
 acquiring a matrix forming a certain matrix equation;   dividing the matrix into a plurality of blocks each including an element having a certain attribute based on an attribute of each element of a plurality of elements included in the matrix;   changing a scale of each element of the plurality of elements included in each block of the plurality of blocks; and   setting the matrix after the scale is changed to a target to be inputted to a machine learning model that performs a matrix operation.   
     
     
         2 . The non-transitory computer-readable recording medium according to  claim 1 , wherein
 the acquiring includes acquiring a first matrix representing a coefficient of the certain matrix equation and a second matrix representing a product of the first matrix and a third matrix representing a solution to the certain matrix equation.   
     
     
         3 . The non-transitory computer-readable recording medium according to  claim 1 , wherein
 the acquiring includes acquiring a third matrix representing a solution to the certain matrix equation.   
     
     
         4 . The non-transitory computer-readable recording medium according to  claim 1 , the process further comprising
 identifying a fourth matrix representing a cardinal number forming each element of the plurality of elements included in the matrix and a fifth matrix representing an index forming each element of the plurality of elements included in the same matrix, wherein   the setting includes setting the fourth matrix and the fifth matrix to a target to be inputted to the machine learning model.   
     
     
         5 . The non-transitory computer-readable recording medium according to  claim 1 , wherein
 the certain matrix equation is a matrix equation to be solved by an iterative solution technique, and   the setting is executed by the computer before the certain matrix equation is solved by the iterative solution technique, wherein   the process further comprising   setting the second matrix representing a solution to the certain matrix equation outputted from the machine learning model to an initial value to be used for solving the certain matrix equation by the iterative solution technique in response to input of the target to be inputted to the machine learning model.   
     
     
         6 . The non-transitory computer-readable recording medium according to  claim 1 , wherein
 the certain matrix equation is a matrix equation to be solved by an iterative solution technique, and   the setting is executed by the computer after the certain matrix equation is solved by the iterative solution technique, wherein   the process further comprising   updating the machine learning model based on the target to be inputted.   
     
     
         7 . The non-transitory computer-readable recording medium according to  claim 5 , wherein
 the acquiring includes acquiring a matrix forming a matrix equation to be solved by the iterative solution technique for the simulation when a plurality of simulations which solve matrix equations of a certain type which are different from each other solved by the iterative solution technique.   
     
     
         8 . The non-transitory computer-readable recording medium according to  claim 1 , wherein
 the certain matrix equation is a matrix equation including a pre-processing matrix to be used for pre-processing including changing an eigenvalue distribution of matrix equations in the iterative solution technique, and   the setting is executed by the computer before the pre-processing is performed by the computer in the iterative solution technique, wherein   the process further comprising   performing the pre-processing based on a third matrix representing a solution to the certain matrix equation which is outputted from the machine learning model in response to the input of the target to be inputted to the machine learning model.   
     
     
         9 . The non-transitory computer-readable recording medium according to of  claim 1 , wherein
 the certain matrix equation is a matrix equation including a pre-processing matrix to be used for pre-processing including changing an eigenvalue distribution of matrix equations in the iterative solution technique, and   the setting is executed by the computer after the pre-processing is performed in the iterative solution technique, wherein   the process further comprising   updating the machine learning model based on the target to be inputted.   
     
     
         10 . The non-transitory computer-readable recording medium according to  claim 8 , wherein
 the acquiring includes acquiring a matrix forming a matrix equation including a pre-processing matrix to be used for pre-processing including changing an eigenvalue distribution of matrix equations in the iterative solution technique for the simulation when a plurality of simulations which solve matrix equations of a certain type which are different from each other solved by the iterative solution technique.   
     
     
         11 . The non-transitory computer-readable recording medium according to  claim 6 , wherein
 the updating includes updating the machine learning model based on the target to be inputted when the machine learning model has solving precision lower than a threshold value.   
     
     
         12 . The non-transitory computer-readable recording medium according to  claim 6 , wherein
 the updating includes updating the machine learning model based on a new target to be inputted, which is acquired by adding a perturbation matrix to the target to be inputted.   
     
     
         13 . The non-transitory computer-readable recording medium according to  claim 2 , wherein the process further comprising
 acquiring a scaling factor acquired by multiplying an inverse of an element having a maximum absolute value in each of the blocks by a certain value, wherein   the changing includes multiplying each element of the plurality of elements included in each block by the scaling factor.   
     
     
         14 . An information processing method for a computer to execute a process comprising:
 acquiring a matrix forming a certain matrix equation;   dividing the matrix into a plurality of blocks each including an element having a certain attribute based on an attribute of each element of a plurality of elements included in the matrix;   changing a scale of each element of the plurality of elements included in each block of the plurality of blocks; and   setting the matrix after the scale is changed to a target to be inputted to a machine learning model that performs a matrix operation.   
     
     
         15 . An information processing device comprising:
 one or more memories; and   one or more processors coupled to the one or more memories and the one or more processors configured to   acquire a matrix forming a certain matrix equation;   divide the matrix into a plurality of blocks each including an element having a certain attribute based on an attribute of each element of a plurality of elements included in the matrix;   change a scale of each element of the plurality of elements included in each block of the plurality of blocks; and   set the matrix after the scale is changed to a target to be inputted to a machine learning model that performs a matrix operation.